AI is starting to help users execute the entire journey "from product search to decision-making to final checkout"! Goldman Sachs breaks down a $2.6 trillion consumer opportunity and identifies 18 potential winners.

Goldman Sachs, the Wall Street financial giant, has released its latest AI agent commerce research report, showing that the new AI agent business model (i.e., Agentic Commerce) fully ignited by Meta Muse and OpenAI Astra is expected to drive a new wave of e-commerce penetration and significantly expand and redistribute commercial value across product discovery, the digital advertising ecosystem, digital payments, and online transaction security. As increasingly consumer-focused personal agents like Muse begin to search for products, compare prices, and execute purchases on behalf of users, the competitive focus in the agent business is shifting to who can capture consumer purchase intent and convert it into trustworthy, fulfillable transactions. Goldman Sachs' latest estimates show that, based on existing consumption scale, AI agent commerce in the U.S. could tap into a potential annual consumer spending pool of approximately $2.6 trillion in the coming years, covering the high-adoption-probability retail and service categories it has identified; growth opportunities mainly come from agents gradually taking over existing shopping expenditures. Goldman Sachs analysts have newly identified approximately $2.6 trillion in existing U.S. consumer spending as belonging to categories relatively easy for agent commerce to adoptthat is, Goldman Sachs expects the existing annual consumer spending pool that agent commerce can capture to be as high as $2.6 trillion, and it is particularly bullish on consumer gateway platforms such as Meta, the parent company of Facebook and Instagram, and Alphabet, the parent company of Google; retail and merchant-type infrastructure such as Amazon, Walmart, and Shopify; and Visa, Mastercard, Cloudflare, and important identity and risk-control service providers. In this research report, Goldman Sachs lists a total of 18 publicly listed companies benefiting from the theme of a massive $2.6 trillion consumer migration. The following ratings and 12-month target prices are based on the report's standards at the time, with reference stock prices as of September 23, 2026, in U.S. dollars; potential upside/downside is calculated based on that reference price. For global equity market investors, the investment opportunity therefore covers both ends: "capturing transactions" and "securing transactions." Goldman Sachs emphasizes that this is a structural change unfolding gradually over the next 35 years or more, and broader adoption may take a longer cycle; companies that can sustainably benefit need to possess user distribution, consumer trust, merchant participation, and transaction execution capabilities. Muse's application deployment and Astra's multi-step execution capabilities provide the technical foundation for further development of this trend, while continued strong demand for CPU, storage, and data center AI infrastructure hardware broadly is the long-term trillion-dollar positive impact on the AI computing power supply chain that can continue to be inferred as application scale expands. From shopping gateway to transaction foundation: Goldman Sachs reveals six paths reshaping the trajectory of online consumer commerce profit distribution First, Goldman Sachs says agent commerce first captures existing consumer spending, then drives e-commerce growth by reducing transaction friction. Third-party surveys cited by Goldman Sachs show that about 44% of online shoppers have already used AI technology to discover products; in another survey, 42% of consumers used AI to compare prices, while only 16% used it to complete purchases, indicating that there is still substantial room for development between product research and actual transactions. Goldman Sachs groups existing U.S. retail and service spending by adoption probability: high-probability categories total about $2.6 trillion, and adding medium-probability categories brings the total to about $12.3 trillion, or about 60% of related spending. Its sensitivity estimates show that converting about 2.3% of offline face-to-face payment spending in high-probability categories to e-commerce could accelerate e-commerce growth by about 1 percentage point; if the denominator is expanded to high- and medium-probability categories, the corresponding conversion ratio is about 0.5%. These are scenario estimates of migration from existing consumption channels and cannot be regarded as newly added $12.3 trillion in market revenue. Adoption speed depends on the subjectivity of product preferences, transaction complexity, and losses from buying the wrong item, so standardized, low-risk purchases are more likely to spread first, while high-end apparel, luxury goods, and other areas may be slower. Second, advertising value will migrate with purchase intent, and Meta and Alphabet have advantages in competing for the new gateway. Recently, AI has already been reducing advertising creative costs, optimizing placement, and improving advertising return on spend; the longer-term change is that consumers may express purchase needs directly in the agent interface, gradually shifting commercial intent from search results pages and merchant websites to AI platforms. Goldman Sachs believes this is similar to the migration from desktop internet to mobile internet: merchants or brands consistently prioritized by agents may gain more stable repeat-purchase relationships. Advertising budgets will also gradually flow toward AI-native sponsored recommendations, product feeds, and other commercial display formats. Meta, with its app ecosystem, user relationships, and Muse, and Alphabet, with Gemini, AI Overviews, AI Mode, and full-stack AI infrastructure, are listed by Goldman Sachs as the main long-term beneficiaries in digital advertising. However, value realization still requires solving recommendation attribution and performance measurement: brands must know whether advertising truly influenced agent selection and the final transaction. Goldman Sachs also emphasizes that, so far, no material impact from agent commerce on retail media performance has been observed. Third, product discovery is becoming more open, potentially benefiting both infrastructure providers for small and medium-sized merchants and large retailers with scale advantages. Goldman Sachs' analyst team says Shopify's opportunity lies in the fact that even as consumer shopping gateways continue to change, merchants still need unified management of product catalogs, inventory, payments, orders, fulfillment, and after-sales service; its UCP and Agentic Storefronts initiatives with Google help merchants connect to multiple AI channels. Amazon and Walmart, meanwhile, have advantages in price, supply, delivery speed, and transaction reliability, which are exactly the metrics agents can directly measure when comparing products. Both models can therefore benefit simultaneously. Goldman Sachs says the divergence in competition lies in who owns the customer relationshipShopify actively connects to new traffic, while Amazon restricts unauthorized shopping agents, reflecting the latter's consideration of protecting first-party data, membership relationships, and retail media revenue. Consumer brands will also diverge: standardized goods are more vulnerable to price comparison and private-label substitution; personalized, high-engagement brands such as Este Lauder are relatively more resilient, and Goldman Sachs also values the brand-building and technological adaptability of SharkNinja and Tapestry. Brand marketing will gradually increase optimization for generative search and agent recommendations, namely the GEO marketing model. Fourth, agent payments will be built on the existing bank card system, and payment networks are expected to gain both transaction volume and incremental service revenue. Goldman Sachs believes Visa and Mastercard are favorably positioned due to their network effects, consumer payment habits, tokenization, and risk-control capabilities. If an agent splits a shopping basket across multiple merchants, the increase in transaction count and decline in average order value may increase revenue per transaction from per-transaction fees; online transactions will also raise attach rates for identity authentication, anti-fraud, and other value-added services. Goldman Sachs says the technical threshold in payment processing will also rise, which it believes benefits native e-commerce processing platforms such as Stripe and Adyen; PayPal's two-sided ecosystem has near-term strategic value, but its competitive position still depends on the actual value it provides to consumers and merchants. Financing may benefit Affirm and Klarna: agents can compare financing costs and terms at purchase, increasing opportunities for buy-now-pay-later options to be discovered and used, but the final payment choice is still determined by consumer preference. Fifth, the more automated the transaction, the more identity, authorization, and liability determination become necessary conditions for commercialization. The technology system is being built in layers: MCP helps connect data and tools, UCP and ACP standardize commercial interactions between merchants and agents, and payment protocols handle identity, authorization, and payment credentials. True scale requires proving "which consumer authorized which agent, to buy what, and with what budget," and clarifying liability for erroneous purchases, fraud, chargebacks, and returns. Traditional IP addresses, device, and browsing behavior signals may decline in independent judgment capability when legitimate agents also operate at high speed through the cloud, thus creating new demand for Cloudflare, Akamai, and payment security service providers. Goldman Sachs estimates that related cybersecurity spending currently equals about 1%2% of U.S. e-commerce revenue, with room to rise over the long term. In information services, Equifax covers consumer, merchant, income, and employment verification; TransUnion excels at continuous identity linkage and digital network risk; FICO's clearer opportunity comes from automated decision-making and anti-fraud software. Goldman Sachs views the agent opportunity for the latter three as long-term growth potential, not yet quantified as clear earnings increments; the key is whether new workflows generate more paid verification and decision-making. Sixth, ticketing-sector survey data cited by Goldman Sachs shows that agents can improve sales efficiency for existing supply, but unique supply still determines platform pricing power. Goldman Sachs cites Live Nation data showing that about 95% of concerts do not sell out, with amphitheater ticket sales rates around 60%70% and theater around 65%75%. If agents can match events based on city, time, budget, and personal preferences, they have an opportunity to increase ticket sales rates and venue utilization for long-tail performances. Goldman Sachs says Live Nation's Ticketmaster is relatively better positioned because it controls differentiated ticket sources and venue partnerships. Secondary ticketing platforms such as StubHub may benefit first from lower customer acquisition costs, but they also face price-comparison transparency, service-fee competition, and reduced ancillary services and advertising sales after consumers bypass original pages. Goldman Sachs still maintains a positive "Buy" rating on both companies, but notes that long-term profit distribution depends on whether agent platforms continue to reduce distribution costs or gradually share more transaction revenue through referral fees, commissions, and display charges. Every AI agent transaction "completed on someone's behalf" may add new AI inference computing workloads From the underlying technology paradigm, Muse and Astra are indeed expected to further expand the scope of agent use, but application penetration will accelerate gradually along different tasks. Muse already has a dedicated cloud virtual machine, browser operation, continuous background work, and memory mechanisms; Astra strengthens computer operation, software use, and multi-step professional task execution capabilities. Applying these capabilities to shopping, a single need may trigger product search, specification verification, inventory checks, price comparison, delivery evaluation, identity verification, and payment confirmation. Higher AI inference workloads, ultimately higher success rates, and lower manual intervention costs will make more daily tasks worth delegating to AI; user scale, usage frequency, and task coverage may thus expand together. At the same time, the account connections, merchant participation, and liability rules emphasized by Goldman Sachs will also determine how quickly technical capabilities real transaction volume. From the underlying architecture of AI computing infrastructure, this expansion of inference workloads will simultaneously increase demand for model inference, tool execution, and state management. Accelerators such as GPUs and TPUs handle model computation; CPUs run browsers, virtual machines, product search, database queries, and transaction orchestration; HBM and server DRAM carry model data, context, and concurrent working environments; enterprise SSDs store product indexes, task records, and persistent state, while high-speed networking and optical interconnects support distributed data exchange. Nvidia's latest engineering materials have clearly pointed out that CPU execution speed affects wait times between agent model calls, and discuss layered management of context caching through HBM, DRAM, local NVMe, and remote storage. This provides a strong engineering basis for long-term demand for x86 architecture CPUs led by AMD and Intel, high-performance Arm architecture CPUs, memory chip components, data center high-speed optical interconnects, and the power supply chain. Undoubtedly, memory chip components for AI data center server clusters remain the clearest supply bottleneck at the AI computing power supply chain level. Market research firm TrendForce expects server DRAM contract prices to cumulatively rise about 270% in 2026, and enterprise SSD prices to cumulatively rise about 235%; in 2027, HBM contract prices may still rise 70%140%. These figures reflect the combined effect of AI computing power expansion and memory price increases. TrendForce's latest estimate shows that DRAM and NAND combined will account for 47% of major cloud service providers' capital expenditure in 2026, rising to 68% in 2027, driven by both higher procurement volumes and higher prices.